Understanding neighbourhood perceptions of alcohol related

anti-social behaviour

Joanna Taylor, Liz Twigg and John Mohan

Abstract

Negative perceptions of anti-social behaviour have been shown by previous research to have

harmful repercussions to both an individual's mental and physical health as well as the

neighbourhood's long term prospects. Studies in the US have previously found that the

location of alcohol supply points is associated with these negative perceptions, whereas

recent, more qualitative and ethnographic research from the UK, emphasises the

heterogonous and contingent nature of attitudes and perceptions towards alcohol consumption

patterns and behaviour.

Using multilevel models applied to data from a national crime survey and geocoded data on

pubs, bars and nightclubs, this paper focuses on the complex relationship between

perceptions of alcohol related anti-social behaviour and the density of such establishments

across England. The findings support the general link between unfavourable perceptions and

density of outlets but also highlight the complexity of this association by showing that these

relationships are dependent on other characteristics of the neighbourhood, namely deprivation

and the proportion of young people in the neighbourhood.

Introduction

Robert Sampson recently stated that “perceptions of disorder constitute a fundamental

dimension of inequality at the neighbourhood level and perhaps beyond” (Sampson, 2009, p.

6) and stressed that it is perceptions of disorder (not observed disorder per se) which has a

crucial influence in social differentiation of urban neighbourhoods. Where residents (and

potential future residents) perceive disorder as an issue there is greater risk that the

neighbourhood will suffer from a downward trajectory (Wikström, 2009). Perceptions of

disorder are also associated with harmful repercussions for an individual’s health (both

physical and mental) and personal well-being (see for example Aneshensel and Sucoff, 1996;

Bowling et al., 2006; Ellaway et al., 2009; Ewart and Suchday, 2002; Steptoe and Feldman,

2001). Understanding the drivers of such perceptions is therefore important to address

neighbourhood inequalities.

In the US, research on neighbourhood perceptions has been led by the Project on Human

Development in Chicago Neighbourhoods (see e.g. Sampson and Raudenbush, 2004, 324). In

the UK, studies have predominantly been based on the Crime Survey for England and Wales

(CSEW), formerly known as the British Crime Survey. This is a large annual household

survey of circa 46,000 individuals (Chaplin et al., 2011) and asks respondents about how

much of a problem they perceive a range of different types of anti-social behaviour (ASB) to

be in their local area.i

Unlike the US research, where disorder is often divided into social and physical disorder (see

e.g. Sampson and Raudenbush, 2004, p324; Skogan, 1990, p. 51-52; Taylor, 2001, p. 56),

perceptions of different types of ASB have been combined for analysis purposes into one

overall indicator (e.g. Kershaw and Tseloni, 2005; Tseloni, 2007; Flatley et al., 2008 and

Taylor et al., 2010). However, more recently this approach has been challenged, warning

against a “reductionist” and “oversimplified” representation of anti-social behaviour (Case et

al., 2011, p. 168). In acknowledgement of this criticism, we focus on one specific dimension

of anti-social behaviour covered by the CSEW, namely “people being drunk or rowdy in

public places”, referred to throughout this paper as “alcohol-related anti-social behaviour”.

However, a second more important substantive reason for focusing attention on this specific

dimension of anti-social behaviour is its importance in the growing debate and discussion

concerning the social implications of the diverse set of alcohol related night time

entertainment and activities that are characteristic of recently regenerated and gentrified

urban areas (Chatterton, 2002, Jayne et al., 2006; 2008). Such spaces of ‘drinkatainment’ are

often linked with anti-social behaviour (Jayne et al, 2006; Bell and Binnie, 2005) and a

burgeoning sense of moral panic (Crawford and Flint, 2009). Debate often stresses the

complexity of this night time economy, highlighting the rivalry between ‘law and order’ and

‘economic profit and growth’ (Chatterton, 2002). The work also emphasises the growth of

flexible and multi-agency governance (Crawford and Flint, 2009) striving to achieve social

control via manipulation of consumption and behaviour patterns. Within all of this fast-paced,

night-time activity centred on alcohol, Hubbard (2013) argues that the ‘practices and

aesthetics of excess’ are an integral part of the ‘urban spectacle’ but may perpetuate divisions

based on race, class and gender. Moreover, others have challenged the widely held

assumption that drunkenness is a transgressive practice both historically (Kneale, 2001) and

in the contemporary city (Latham, 2003).

Much of this established and emerging literature on the urban nightscape is heavily

theoretical and often based on ethnographic and case study research. In contrast, the work

presented here is heavily quantitative but attempts to supplement this rich literature by

summarising the underpinning complexity of perceptions of anti-social behaviour relating to

alcohol across the whole of England.

Background: perceptions of alcohol related anti-social behaviour and alcohol supply

points

Evidence for England suggests that there are strong geographical variations in unfavourable

perceptions of alcohol-related ASB, ranging from 17 per cent in North Yorkshire to 35 per

cent in Greater Manchester (Walker et al., 2009). However, the mechanisms by which

alcohol use affects perceptions of local neighbourhood are multi-faceted and complex, and

arguably, reflect the diversity and heterogeneity of drinking cultures, practices, behavioural

norms, urban design and nightscape activities outlined in the literature above (but see Jayne

et al, 2008 and Moon and Kearns, 2014, p235 for reviews with a geographical focus). The

research which explicitly centres on explaining disparities has tended to be piecemeal,

usually focused in one region or place, although some work is based on national studies.

Flint et al. (2007), working with Glasgow Housing Association, for example, found that

tenants made reference to the perceived environmental degradation of the local area, both in

terms of groups of people consuming alcohol in public places and through broken bottles and

associated litter around buildings. In Cardiff and Swansea (Bromley et al., 2000), people’s

perceived insecurity was found to be highly localised and related to crime and ASB

associated with the night-time economy. Focusing on the London Boroughs, Brunton-Smith

et al. (2010) showed that area level deprivation was significantly associated with individual

negative perceptions of alcohol related ASB, independent of the type of person interviewed.

In an extensive study working with the 2007/8 sweep of the Crime Survey for England and

Wales, Flatley et al. (2008) found that the frequency with which survey respondents visited

the pub was found to be related to negative perceptions of alcohol-related ASB.

Some studies have focused specifically on alcohol outlet density. Outside of the UK

framework, evidence from New South Wales in Australia has shown that such density is

associated with perceptions of neighbourhood problems relating to drunkenness (Donnelly et

al., 2006). This research also indicated that those living in more deprived areas as well as

younger residents were also more likely to report adverse perceptions. Sampson and

Raudenbush’s (2004) study of Chicago used systematic social observation of land use which

measured incidents of commercial building security (such as iron security gates or pull down

metal shutters), alcohol and tobacco advertising and bar and liquor stores which, in

combination, they treated as a measure of alcohol density. Although the measure has been

criticised for not focusing solely on alcohol-related land use (see McCord et al., 2007), it was

found to be statistically significantly related to perceptions of both physical and social

disorder.

In summary, the evidence from localised studies in the UK and the seminal US study of

Chicago suggest that availability (ie density) of alcohol outlets is, in some way, associated

with unfavourable perceptions of generic or alcohol related anti-social behaviour. Findings

also suggest however, that the relationship is contingent on individual characteristics as well

as features of neighbourhood. Up until recently it was not possible to explore these

relationships across the small areas of England because of data limitations. Whilst the CSEW

was able to supply information on alcohol related ASB, it did not contain neighbourhood

counts of alcohol outlets. However, as we outline below, changes in the way that these data

are now made available allow us to link the individual level survey findings to external

sources on alcohol outlets for neighbourhoods. This data structure facilitates a multilevel

analysis whereby we can investigate the extent to which spatial disparities in alcohol related

ASB are associated with individual characteristics (e.g. age, gender and socio-economic

background) and area characteristics (including density of alcohol outlets) simultaneously. In

this way we can explore some of the more nuanced debates regarding the influence of high

density alcohol related activities in urban space on perceptions of alcohol related ASB.

At this stage it is useful to specify the research questions that the paper will explore by means

of a series of multilevel models (referred to throughout this paper as Models A to D). In the

first model (A) we are interested in the individual characteristics associated with negative

perceptions of alcohol-related ASB. We then extend this model to test which area

characteristics influence such perceptions after controlling for individual compositional

effects (Model B). By adding the density of pubs, bars and nightclubs to the model we

quantitatively test the intuitive hypothesis that higher densities will be related to more

negative perceptions above and beyond other features of the respondent’s local area (Model

C). In Model D, we test the possible mediating effects of the density of pubs, bars and

nightclubs in the relationship between neighbourhood deprivation, levels of young people and

perceptions of alcohol-related ASB (Model D).

Data sources

Our main data source was the 2008/9 sweep of the Crime Survey for England and Wales.

Although the CSEW covers England and Wales, one of our key independent variables – the

2010 Index of Multiple Deprivation – was not available for Wales and therefore our analysis

and results are restricted to England only. Full technical details of the survey can be found in

Bolling et al., (2009). The analysis presented here takes advantage of a recent innovation

whereby the Lower Layer Super Output area identification code for each observation has

been attached to the dataset. This areal unit relates to UK Census Geography and has a mean

population of 1,500. For a fuller description see (Office for National Statistics, 2012).

Importantly, the inclusion of this spatial identifier allows us to link the CSEW survey results

to external area level data sources such as the 2001 UK Census (Office for National Statistics,

2004a) and Ordnance Survey’s MasterMap Address Layer 2 dataset (Ordnance Survey,

2011a). For the purposes of this study we worked at the spatial level of Middle Super Output

Area (MSOA) which are derived from aggregations of Lower Super Output Areas and have a

mean population of 7,200. Arguably, MSOAs approximate to the definition of local area

given to CSEW interviewees (a 15 minute walk from the respondent’s home) and have been

used elsewhere with CSEW data to define local neighbourhoods (see for example Brunton-

Smith and Sturgis, 2011).

Dependent variable

Our main dependent variable is the CSEW question “how much of a problem are people

being drunk or rowdy in public places in your local area?”, with the possible answers being

“a very big problem”, “a fairly big problem”, “not a very big problem” or “not a problem at

all”. Respondents are not prompted to think about a particular day of the week or time of day

when formulating their answers, although it is highly feasible that actual levels of alcohol-

related ASB could differ. Those answering either of the first two responses were categorised

as perceiving problematic levels of alcohol related ASB.

Independent variables

We were interested in identifying the covariates of perceptions of alcohol-related ASB at the

individual, household and area level. Because of the limited sources of research into the

different dimensions of ASB (Brunton-Smith et al., 2010; Flatley et al., 2008), our variable

selection was also guided on overall perceptions of ASB (Franzini et al., 2008; Millie et al.,

2005; Sampson and Raudenbush, 2004; Taylor et al., 2010). Although the main focus of this

paper is on area context (ie density of outlets), it is important to acknowledge the potential

impact of an individual’s socio-economic and demographic make-up in forming individual

perceptions. For this reason, and guided by the literature, individual factors, namely gender,

age, ethnicity, marital status, educational qualifications and whether the person had been a

recent victim of crime were all included in the models. Important household level

characteristics were also included (i.e. income, tenure, accommodation type and length of

time living in the neighbourhood). Measuring place effects was also informed by the wider

body of literature on perceptions towards an overall or combined measure of ASB. Shaw and

McKay’s (1942) classic study of Chicago found that neighbourhoods with high levels of

disorder were more likely to experience low-socio economic status, high population turnover

and high levels of ethnic heterogeneity. To capture low socio-economic status

neighbourhoods we employed the 2010 Index of Multiple Deprivation (DCLG, 2011).ii This

index combines seven domains, chosen to cover a range of economic, social and housing

issues. The living environment domain consists of two sub-domains which attempt to identify

deprivation in the quality of the local environment both within and external to the home.

(McLennan et al., 2011). Because of the circular relationship between our dependent variable

and this external assessment of environment we decided to exclude this sub-domain from the

overall deprivation score. The crime domain was also removed as level of reported crime is

included in our models as a separate independent variable. The remaining domains (including

the ‘indoor’ living environment, quantified in the IMD as houses in poor condition and/or

without central heating), were then merged into a composite score which we have labelled

‘deprivation’ in our models.

Population turnover was derived from MSOA inflow and outflow rates per 1,000 population

for the period July 2008 to June 2009 (Office for National Statistics, 2010a). Because these

two variables were highly correlated (0.94) they were combined using Principal Component

Analysis into one variable to represent turnover between MSOAs.

To measure ethnic diversity, we employed the Gibbs and Martin (1962) index defined as :

∑

where p = proportion of individuals in a category (from the 2001 Census) and N = number of

categories. The range of the index of diversity is from 0 (where all residents are of the same

ethnic background) to ⁄ or in this case 0.8 where all five Census ethnic groups (white,

mixed, black or black British, Asian or Asian British and Chinese or other) have equal

representation in a neighbourhood. Ethnic heterogeneity, alongside all the other continuous

area variables, were standardised (with mean=0 and standard deviation=1).

Following Sampson and Raudenbush (2004) discussion of other important area influences we

also include four other area level variables. First is the level of young people (defined here as

those aged 15 to 24 years old) in the population (Office for National Statistics, 2010b) and

second is a cross-government rural and urban area classification which divides

neighbourhoods into three types; ‘urban’ (defined as urban settlements with a population

greater than 10,000), ‘small town and fringe’ and ‘village, hamlet and isolated dwellings’

(The Countryside Agency et al., 2004). The third variable is the reported incidents of anti-

social behaviour and in the absence of systematic social observation data (as used in the

Chicago studies), police data on levels of actual anti-social behaviour have been used. Since

December 2010 street level information on crimes reported to the police has been available at

www.police.ukiii

. These point level reported crime data were imported into a Geographical

Information Software package, ArcGIS v10 (ESRI, 2011) along with digital boundary data

for MSOAs (Office for National Statistics, 2004b). The software was then used to calculate a

count of incidents of anti-social behaviour reported to the police per 1,000 population for

each neighbourhood. It is important at this juncture to highlight the fact that these data will

only capture a proportion of actual incidents of ASB. Indeed the 2007/08 sweep of the CSEW

found that 72 per cent of individuals who witnessed any type of anti-social behaviour did not

report it to the police or any other authority (Innes and Weston, 2010).

Most crucially to the research objectives of this study, the fourth item measures the density of

pubs, bars and nightclubs in the respondent’s local area. The decision to focus this research

on pubs, bars and nightclubs was two-fold. First, other alcohol supply points such as off-

licences are not separately identified in the Ordnance Survey’s MasterMap Address Layer 2

database. Secondly previous research has shown that it is this category of alcohol supply

points which are the strongest predictor of variations in crime (Newton et al., 2010).

However, as Newton and Hirschfield (2009) reported there is no single source of consistent

data on these supply points in England and Wales. We therefore used the Ordnance Survey’s

MasterMap Address Layer 2 database which contains coordinates for more than 27 million

residential and commercial properties in Great Britain. The database employs three different

classification schemes to denote the nature of buildings– the Ordnance Survey Base Function,

the National Land Use Database Group and the Valuation Office Agency non-domestic rates

primary description code (Ordnance Survey, 2011b) however only the first of these three

coding schemes offers complete coverage. Further, there is a substantial proportion of

commercial premises that are simply coded as “general commercial”. Consequently we are

unable to ascertain whether these buildings are pubs, bars or nightclubs. This is not an

unsubstantial problem when using the data. For example Smith and Crooks (2010) reported

that in London 51 per cent of non-residential addresses had the “general commercial

classification”, leading them to conclude that MasterMap Address Layer 2 building

functionality should not be used in detailed analysis without being aware of the errors in the

commercial classifications. However, as there is no obvious reason to suggest that the use of

the general commercial ‘catch-all’ category is unevenly employed between business types

and in absence of an alternative dataset of alcohol supply points covering the whole of

England we have used Address Layer 2 here while fully acknowledging the dataset’s

shortcomings. By using the combination of land use classifications we were able to calculate

the density of pubs, bars and nightclubs across the MSOAs.iv

Analytical approach

To investigate the research questions set out in the background section above, we employed

multilevel logistic modelling (Goldstein 2003; Snijders and Bosker 1999). Such models

adjust standard errors for the clustered sample design used in the CSEW, whereby individuals

(n=41,892) are nested within MSOAs (n=3,611), which in turn are nested within Police Force

Areas (n=38).v These models are also appropriate for distinguishing between the variance

associated with individuals and the variance relating to the different geographical levels

within the data structure and thus avoid incorrect conclusions based on ecological or

individualistic fallacies.

Each of the models has a single binary response variable (see section on the dependent

variable above) and were initially estimated using iterative generalised least squares based on

a first order marginal quasi-likelihood approximation using the software package MLwiN

(Rasbash et al., 2009). The model coefficients were checked for stability using Markov

Chain Monte Carlo simulation (Browne, 2009a). The models were each run through 50,000

iterations (with a burn-in period of 5,000 iterations). The first model (A) explored whether

individuals in similar environments perceived different levels of alcohol-related ASB. To do

this, the first model contained our pertinent individual level variables, but to estimate how

these variables are associated with perceptions of alcohol related ASB within MSOAs, each

variable was centred around its MSOA mean (see Kawachi and Subramanian, 2006; Sampson

and Raudenbush, 2004). The subsequent models (B to D) involve the assessment of area

level covariates on individual perceptions of alcohol related ASB. Here all individual level

variables are centred around their grand mean which ensures that area level associations will

be adjusted for individual-level characteristics. In all models we allowed random intercepts

but not random slopes.

Results

Table 1 shows the results for Model A which focuses on the association between individual

level characteristics and perceptions of alcohol related ASB. Results are expressed as logits.

Credible intervals, derived via the Bayesian estimation process, which can be interpreted in

much the same way as confidence intervals, are also included. The results illustrate that in

line with previous research on perceptions of alcohol-related ASB (Brunton-Smith et al.,

2010; Donnelly et al., 2006), as a person gets older they are less likely to perceive alcohol

related ASB to be a problem. As Egan et al. (2012) noted these findings run contrary to the

direction one would expect to see if negative perceptions of anti-social behaviour were driven

by the UK’s intergenerational intolerance and characterisation of younger people as anti-

social (as argued by the United Nations Committee on the Rights of the Child (2008)) – a

stereotype which young people themselves identify with (Neary et al., 2013). Possible

explanations could be that older people, on average, live further away from any night time

economy within their neighbourhood and/or interact less with those areas where alcohol

activities are concentrated. In university towns and cities, the ‘town’ versus ‘gown’ conflict

(Hall, 1997; Munro and Livingston, 2012), for example, may lead urban planners, and pub

landlords to create entertainment and accommodation areas that are deliberately exclusionary

to non-students.

Table 1 Individual and household level factors associated with negative

perceptions of alcohol related anti-social behaviour (model A)

β Credible interval

Gender (base=female) Male -0.08 -0.13 -0.03

Age -0.02 -0.02 -0.02

Ethnicity (base=white) Mixed 0.26 -0.05 0.56 Asian or Asian British 0.10 -0.05 0.24 Black or Black British -0.21 -0.39 -0.03 Chinese or Other 0.34 0.10 0.58

Marital status (base=married) Single 0.08 0.01 0.16 Widowed -0.19 -0.31 -0.08 Separated or divorced 0.13 0.04 0.22

Victim of CSEW crime in past 12 months (base=non victim)

Victim of CSEW crime 0.78 0.72 0.84

Educational qualifications (base=below A level or none) A level or above 0.03 -0.02 0.10

Household income (base=£40k plus) Under £5k 0.25 0.11 0.39 Under £10k 0.18 0.05 0.30 Under £20k 0.19 0.10 0.29 Under £30k 0.23 0.14 0.32 Under £40k 0.14 0.05 0.24 Don’t know or refused income 0.08 -0.01 0.18

Tenure (base=owner occupier) Social rented sector 0.21 0.13 0.30 Private rented sector 0.08 -0.01 0.16

Accommodation type (base=house) Flat/maisonette/bedsit 0.23 0.14 0.33 Other accommodation (including not coded) 0.15 -0.01 0.31

Time living in neighbourhood (base=five years or more) Less than 12 months -0.57 -0.69 -0.45 Less than 5 years -0.30 -0.37 -0.23

Notes:

1. The weighting variables number of adults in the household (nselec) and number of households at the address

(hselec) were also include. β(nselec)=0.06 (with a credible interval of 0.03 to 0.10) and β(hselec)=0.01 (credible

interval -0.06 to 0.07). As stipulated by Rabe-Hesketh and Skrondal (2006) models were also produced without

these two design variables to test whether these extra covariates altered the interpretation of the regression

coefficients of interest – overall they did not with the exceptions of marital status where being β(single)=0.06

(with a credible interval of -0.02 to 0.13) and β(separated or divorced)=0.08 (credible interval -0.01 to 0.16) and

(ii) tenure where β(private rented sector)=0.09 (credible interval 0.01 to 0.18).

2. “0.00” indicates .

Being male, widowed or living in the area for a short period of time all reduce the chances of

perceiving drunk or rowdy behaviour to be a problem in the local area. Conversely

respondents who described themselves to the survey as being either Asian or Asian British or

Chinese (or from another minority ethnic background) were more likely to perceive a

problem. Those in the lower income brackets (household incomes of less than £30k per

annum) as well as those living in flats or in social rented housing were also more likely to

perceive a problem. There are many number of reasons why poorer economic status may lead

to more adverse perceptions. It may be that such people live in less gentrified areas of the

city where the night time economy is characterised by fewer numbers of security personnel,

compared to more up-market areas (see, for example, Chatterton, 2002).

The impact of previous crime victimisation was particularly strong; if a respondent had been

a recent victim of crime, the odds of them describing alcohol related ASB to be a problem

more than doubled. No association was found between educational achievement and

perceptions of alcohol-related ASB.

The second stage of the modelling focuses on the area-level variables. In these models (Table

2), estimates of the coefficients of the area factors have been adjusted for the characteristics

of individuals and households as detailed in Table 1.vi

Unfailingly researchers have found a

link between neighbourhood deprivation and negative perceptions of disorder on both sides

of the Atlantic – the results presented here being no exception (Model B). The proposition

that there is a link between ethnic density and/or levels of ethnic diversity and perceptions of

anti-social behaviour has been controversial (for a full discussion see Taylor et al., 2010).

The results here suggest that residents who live in an ethnically mixed area were less likely to

report alcohol related anti-social behaviour to be a problem in their neighbourhood.

Table 2 Area level factors associated with negative perceptions of alcohol related anti-social behaviour (models B to D)

Model B – area effects Model C – pubs etc. added

to the model Model D – moderating

effects

β Credible interval

β

Credible interval

β

Credible interval

Deprivation 0.14 0.10 0.19

0.16 0.11 0.20

0.15 0.10 0.19

In-flow and out-flow between neighbourhoods 0.17 0.12 0.22

0.16 0.11 0.22

0.16 0.10 0.21

Ethnic heterogeneity -0.16 -0.22 -0.11

-0.17 -0.22 -0.11

-0.16 -0.21 -0.10

Proportion of the population aged 15 to 24 0.02 -0.03 0.07

0.02 -0.03 0.06

0.03 -0.02 0.08

Rural and urban area classification (base=urban greater than 10k)

Town and fringe -0.08 -0.21 0.04

-0.08 -0.20 0.04

-0.06 -0.17 0.06

Village, hamlet and isolated dwellings -1.07 -1.22 -0.92

-1.08 -1.22 -0.93

-1.04 -1.19 -0.90

Reported incidents of ASB 0.14 0.10 0.19

0.11 0.06 0.16

0.12 0.07 0.17

Pubs, bars and nightclubs per km2

0.07 0.03 0.11

0.10 0.06 0.15

Pubs * deprivation

-0.05 -0.09 -0.01

Pubs * people aged 15 to 24

-0.03 -0.05 -0.00

Notes:

1. “-0.00” indicates .

Perceptions of drink-associated ASB were worse in neighbourhoods with high levels of

population turnover and more urban areas. As would be expected, Model B indicates that

actual reported levels of anti-social behaviour increased an individual’s propensity to report

high levels of alcohol related anti-social behaviour (regardless of whether they themselves

have been a recent victim of crime). The proportion of young people living in the

neighbourhood did not have an independent effect on perceptions.

Model C shows the effects of the density of pubs, bars and nightclubs on perceptions of

alcohol-related ASB. The results indicate that individuals who live in areas with the highest

density of pubs, bars and nightclubs do perceive more problematic levels of drunk and rowdy

behaviour. Moreover, this relationship holds after adjusting for personal, household and other

area characteristics.

The final stage of the analysis examines whether the effect of alcohol outlet density on

perceptions of ASB is moderated by other factors. Does the contextual influence of pubs and

clubs vary across different types of people and/or different types of places? This is highly

plausible given the complexity of attitudes towards ‘drinkatainment’, the varied nature of the

urban night time economy and the diversity of consumers and providers involved with

alcohol-related activities (Jayne et al, 2008, Chatterton, 2002, Latham, 2003). Are residents

expectations around what is acceptable behaviour, as Millie (2008, 379) contends

“determined by social and cultural norms of aesthetic acceptability”? Model D shows two

interaction terms between the density of pubs and other area level variables namely

deprivation and the level of young people in the local area.vii,viii

The attenuating effects of

both a younger community and deprivation on the density of pubs and clubs are illustrated in

Fig 1. As the neighbourhood either becomes more youthful or more deprived the probability

of negative perceptions of alcohol related ASB becomes the same regardless of the density of

pubs and clubs.

Figure 1 Area level interactions with the density of pubs, bars and nightclubs

It is difficult to hypothesise on the mechanisms behind these findings in a cross sectional

survey although it is arguable that in high deprivation areas or those where many young

people live, pubs are not necessarily seen as a “red-flag” in terms of troublesome behaviour,

they are instead, using Millie’s phraseology, tolerated or even celebrated in some instances as

a positive part of community life. Young people may even influence cultural norms of

acceptance in areas of high youth concentration. They are often the dominant group who

consume the 24 hour city and nearly half of young people in the UK now attend higher

education (Department for Business Innovation and Skills, 2013). They are probably familiar

with extreme alcohol occasions such as Fresher’s week and Carnage drinking events

(Hubbard, 2013) and therefore may be more accepting of behaviour that others find

disdainful.

Again, it is difficult to explain why deprivation attenuates the association with density outlet.

It may be that areas defined as deprived by the Index of Multiple Deprivation may also be

those same parts of inner cities and urban areas that have a thriving, well-managed, night

time economy. In such areas, there may be a high density of outlets but levels of acceptance

are more tolerant and flexible management curtails levels of alcohol related ASB.

It is worth noting that although the final model (Model D) explains 53 per cent of the area

level variation in the modelix

this means a significant proportion of the area level variation

remains unexplained by the area factors described above. A potential explanation of this is

apparent by analysis of additional CSEW questions. Although the majority (86%) of

individuals living in England reported forming their negative perceptions, at least in part,

from their personal experiences, a significant proportion of adults who perceived drunk and

rowdy behaviour to be a problem in their local area recounted forming their opinion based on

factors which are not necessarily specific to their immediate neighbourhood – local

newspaper stories, TV or radio (20%), friends and family (29%) or simply being just

something that is well known (25%). Indeed, of those who reported alcohol-related ASB to

be a problem in their local area, when asked specifically whether they had witnessed drunk or

rowdy behaviour close to where they lived, over a third (37%) said they had not.

Conclusions

A substantive finding of this work is that it provides evidence, for England, that the location

of pubs, bars and nightclubs is associated with negative perceptions of alcohol-specific anti-

social behaviour even after controlling for actual levels of anti-social behaviour reported to

the police. Moreover this contextual relationship varies depending on both the level of

deprivation and the proportion of young people in the neighbourhood. It should be noted that

obviously not all neighbourhoods with a high density of pubs and clubs are necessarily

associated with worsening perceptions; some implement conscientious place management or

strong anticrime policies (McCord et al., 2007). Furthermore, it is important at this juncture

to acknowledge that pubs, bars and nightclubs also vary considerably in terms of their size,

pricing policies, turnover and trading hours (Livingston, 2011) none of which are taken into

account in the analyses presented here.

The results presented here also suggest that previous research which has amalgamated

different types of anti-social behaviour into one overall measure may have masked differing

associations between types of ASB. For example a high proportion of young people in the

neighbourhood has been found to be related to negative perceptions of combined measures

of neighbourhood anti-social behaviour (Ames et al., 2007; Kershaw and Tseloni, 2005;

Taylor et al., 2010; Tseloni, 2007). However, there was no statistically significant main effect

when looking at the more focused question of perceptions of drunk and rowdy behaviour

examined here but there was a significant interaction between levels of young people and

density of outlets. Accordingly, future research plans include investigating differing

associations between perceptions of diverse types of anti-social behaviour in more detail.

Although these results add to previous work by examining the relationship between

perceptions of alcohol related anti-social behaviour and the location of pubs, bars and

nightclubs in an English context, there are several ways in which our analysis might be

refined. Firstly we cannot be certain that our chosen definition of neighbourhoods, MSOAs,

correspond to the area in which respondents live their lives or indeed the area which they

were thinking about when interviewed. Moreover a respondent may live near an MSOA

boundary and be recalling their perceptions of the adjoining MSOA – a problem which

Sampson and Raudenbush (2004, p. 333) referred to as “spatial mismatch”. Nor can we

create bespoke neighbourhoodsx (or indeed know whether respondents live close to a

boundary) as confidentiality restraints (with respect to not disclosing the exact location of

individual respondents) prevented this. Furthermore, even this more complex and resource

intensive approach would not have addressed the flexible notion of neighbourhoods that any

one individual may possess and which are likely to be contingent on life stage or routine

activity (e.g., employment and family status along with leisure pursuits etc). A second

problem with our choice of MSOAs is the modifiable areal unit problem (Openshaw, 1984).

Detail is inevitably lost when contextual data are aggregated to a larger geographical level

and there is no guarantee that adopting an alternative geography would not have resulted in a

different set of findings. Finally a more sophisticated analysis of the distribution of pubs,

clubs and nightclubs such as the network analysis employed by Ellaway et al., (2010) was not

undertaken due to the large geographical area covered by this study.

This research has demonstrated how the development of geocoded social survey datasets will

enable further work using ‘proper’ place specific factors (such as localised crime rates and

building usage) rather than simply relying on aggregates of individual statistics. Newton et

al., (2010, p. 1) previously made the argument that the lack of data on alcohol supply points

“impairs attempts to gain a strategic overview of the timing and location of the availability of

alcohol, the proximity of the various outlets to each other, and their relationship to crime and

disorder” and the results presented here echo their sentiment. In order to fully utilise these

enhanced social survey datasets it is imperative that administrative datasets, such as any

database of licensed premises, include localised geographical information.

Here, the localised information on alcohol outlets and reported incidents of ASB have

allowed the work to move beyond the ‘black box’ nature of area effects. We have attempted

to detail the ways in which place characteristics may influence perceptions of alcohol related

ASB. Here we can draw parallels with the approaches and debates in the public health and

social epidemiology literature concerning the influence of area contexts on health-related

behaviours such as smoking, diet, exercise and alcohol consumption. Traditional approaches

looked at the correlates of these risky behaviours adopting a broad-brush approach usually

encompassing area deprivation scores. Now more nuanced debate argues for a deeper

understanding of the links between context and behaviour, stressing that any identified

protective or harming contextual effects are contingent; that is contextual influences vary

across different types of people and different types of places. Furthermore, contextual

processes are not static; place or context is characterised by dynamic, multi-scalar, socio-

relational complexity (Cummins et al., 2007; Pearce et al., 2012).

Acknowledgements

We would like to thank the four anonymous reviewers for their constructive comments on an

earlier version of this paper. We are grateful to the UK Data Service, Department for

Communities and Local Government, the Office for National Statistics and Ordnance Survey

for allowing us access to the Crime Survey for England and Wales, the 2010 Index of

Multiple Deprivation, results from the UK 2001 Census and MasterMap Address Layer 2

respectively.

Funding

This research was supported by the Economic and Social Research Council [grant number

ES/H019235/1].

Ames, A., Powell, H., Crouch, J. and Tse, D. (2007). Anti social behaviour; people, place

and perceptions. London: Ipsos MORI.

Aneshensel, C. and Sucoff, C. (1996). The neighborhood context of adolescent mental health.

Journal of Health and Social Behavior, 37, pp. 293-310. DOI: 10.2307/2137258.

Bell, D. & Binnie, J. (2005) “What’s eating Manchester? Gastro-culture and urban

regeneration.’ In Franck, K.A. (ed.) Food and The City, Chichester, Wiley.

Bolling, K., Grant, C. and Donovan, J. (2009). 2008-09 British Crime Survey (England and

Wales) technical report volume I. London: BMRB.

Bowling, A., Barber, J., Morris, R. and Ebrahim, S. (2006). Do perceptions of neighbourhood

environment influence health? Baseline findings from a British survey of aging.

Journal of Epidemiology and Community Health, 60, pp. 476-483. DOI:

10.1136/jech.2005.039032.

Bromley, R., Thomas, C. and Millie, A. (2000). Exploring safety concerns in the night-time

city: Revitalising the evening economy. Town Planning Review, 71, pp. 71.

Browne, W. (2009a). MCMC estimation in MLwiN version 2.1. Bristol: Centre for Multilevel

Modelling, University of Bristol.

Brunton-Smith, I., Sindall, K. and Tarling, R. (2010). The effect of demographic make-up on

perceptions of anti-social behaviour in London as measured by the British Crime

Survey and 2008 Place Survey London: Government Office for London.

Brunton-Smith, I. and Sturgis, P. (2011). Do neighbourhoods generate fear of crime? An

empirical test using the British Crime Survey. Criminology, 49, pp. 331-369. DOI:

10.1111/j.1745-9125.2011.00228.x.

Case, S., Ellis, T., Haines, K., Hayden, C., Shalev, K. and Shawyer, A. (2011). A tale of two

cities: Young people, anti-social behaviour and localised public opinion. Crime

Prevention and Community Safety, 13, pp. 153-170. DOI: 10.1057/cpcs.2011.2.

Chaplin, R., Flatley, J. and Smith, K. (2011). Crime in England and Wales 2010/11: Findings

from the British Crime Survey and police recorded crime (First Edition). London:

Home Office.

Chatterton, P (2002). Governing Nightlife: Profit, Fun and (Dis)Order in the Contemporary

City. Entertainment Law, Vol.1, No.2, pp.23–49.

Crawford, A. and Flint, J. (2009). Urban safety, anti-social behaviour and the night-time

economy. Criminology and Criminal Justice, Vol: 9(4): 403–413.

Cummins, S., Curtis, S., Diez-Roux, A. and Macintyre, S. (2007). Understanding and

representing 'place' in health research: A relational approach. Social Science &

Medicine, 65, pp. 1825-1838. DOI: 10.1016/j.socscimed.2007.05.036.

DCLG. (2011). The English Indices of Deprivation 2010. Downloaded from

https://www.gov.uk/government/publications/english-indices-of-deprivation-2010.

Department for Business Innovation and Skills (2013). Participation rates in higher

education, 2006/2007 – 2011/2012 (provisional). London: Department for Business

Innovation and Skills.

Donnelly, N., Poynton, S., Weatherburn, D., Bamford, E. and Nottage, J. (2006). Liquor

outlet concentrations and alcohol-related neighbourhood problems. Sydney: NSW

Bureau of Crime Statistics and Research.

Egan, M., Bond, L., Kearns, A. and Tannahill, C. (2012). Is concern about young people's

anti-social behaviour associated with poor health? Cross-sectional evidence from

residents of deprived urban neighbourhoods. Bmc Public Health, 12, pp. 217.

Ellaway, A., Macdonald, L., Forsyth, A. and Macintyre, S. (2010). The socio-spatial

distribution of alcohol outlets in Glasgow city. Health & Place, 16, pp. 167-172. DOI:

10.1016/j.healthplace.2009.08.007.

Ellaway, A., Morris, G., Curtice, J., Robertson, C., Allardice, G. and Robertson, R. (2009).

Associations between health and different types of environmental incivility: A

Scotland-wide study. Public Health, 123, pp. 708-713. DOI:

10.1016/j.puhe.2009.09.019.

ESRI, 2011. ArcGIS Desktop 10. Redlands, CA: Environmental Systems Research

Institute.

Ewart, C. and Suchday, S. (2002). Discovering how urban poverty and violence affect health:

Development and validation of a neighborhood stress index. Health Psychology, 21, pp.

254-262. DOI: 10.1037//0278-6133.21.3.254.

Flatley, J., Moley, S. and Hoare, J. (2008). Perceptions of anti-social behaviour: Findings

from the 2007/08 British Crime Survey supplementary volume 1 to crime in England

and Wales 2007/08. London: Home Office.

Flint, J., Casey, R., Davidson, E., Pawson, H. and McCoulough, E. (2007). Tackling anti-

social behaviour in Glasgow: An evaluation of policy and practice in the Glasgow

housing association. Glasgow: Glasgow Housing Association.

Franzini, L., Caughy, M., Nettles, S. and O'Campo, P. (2008). Perceptions of disorder:

Contributions of neighborhood characteristics to subjective perceptions of disorder.

Journal of Environmental Psychology, 28, pp. 83-93. DOI:

10.1016/j.jenvp.2007.08.003.

Gibbs, J. and Martin, W. (1962). Urbanization, technology and the division of labor:

International patterns. American Sociological Review, 27, pp. 667-677.

Goldstein, H. (2003). Multilevel Statistical Models. Third Edition. London, Edward Arnold.

Hall, P. (1997) The university and the city, Geo Journal, 41(4), pp. 301–309.

Hubbard, P. (2013) Carnage! Coming to a town near you?

Nightlife, uncivilised behaviour and the carnivalesque body. Leisure Studies Vol. 32,

No. 3, 265–282.

Innes, M. and Weston, N. (2010). Re-thinking the policing of anti-social behaviour. Report

prepared for HMIC by Cardiff University.

Jayne, M., Holloway, S., and Valentine, G. (2006) Drunk and Disorderly: alcohol, urban life

and public space, Progress in Human Geography 30(4): 451-68

Jayne, M., Valentine, G., and Holloway, S. (2008) Geographies of alcohol, drinking and

drunkenness: a review of progress, Progress in Human Geography 32(2) (2008) pp. 247–

263 Johnston, R., Jones, K., Sarker, R., Burgess, S. and Bolster, A. (2004). Party support and the

neighbourhood effect: spatial polarisation of the British electorate, 1991-2001. Political

Geography, 23, pp. 367-402. DOI: 10.1016/j.polgeo.2003.12.008.

Kawachi, I. and Subramanian, S. (2006). Measuring and modeling the social and geographic

context of trauma: a multilevel modeling approach. Journal of Traumatic Stress, 19, pp.

195-203. DOI: 10.1002/jts.20108.

Kershaw, C. and Tseloni, A. (2005). Predicting crime rates, fear and disorder based on area

information: Evidence from the 2000 British Crime Survey. International Review of

Victimology, 12, pp. 293-311.

Kneale, J. (2001) The place of drink: temperance and the public 1956–1914. Social and

Cultural Geography 2(1): 43–59.

Latham, A. (2003) Urbanity, lifestyle and making sense of the new urban cultural economy:

notes from Auckland, New Zealand. Urban Studies 40: 1699–724.

Livingston, M. (2011). Alcohol outlet density and harm: Comparing the impacts on violence

and chronic harms. Drug and Alcohol Review, 30, pp. 515-523. 10.1111/j.1465-

3362.2010.00251.x.

McCord, E., Ratcliffe, J., Garcia, R. and Taylor, R. (2007). Non-residential crime attractors

and generators elevate perceived neighborhood crime and incivilities. Journal of

Research in Crime and Delinquency, 44, pp. 295-320. DOI:

10.1177/0022427807301676.

McLennan, D., Barnes, H., Noble, M., Davies, J., Garratt, E. and Dibben, C. (2011). The

English indices of deprivation 2010. London: Department for Communities and Local

Government.

Millie, A. (2008). Anti-social behaviour, behavioural expectations and an urban aesthetic.

British Journal of Criminology, 48, pp. 379-394. DOI: 10.1093/bjc/azm076.

Millie, A., Jacobson, J., McDonald, E. and Hough, M. (2005). Anti-social behaviour

strategies: Finding a balance. Bristol: Policy Press.

Moon, G. and Kearns, R. (2014). The Sick City: Challenges and Opportunities for Urban

Health. In Ronan Paddison and Eugene McCann (Eds) Cities and Social Change.

(Chapter 9). London: Sage

Munro, N. and Livingston, M. (2012) Student Impacts on Urban Neighbourhoods: Policy

Approaches, Discourses and Dilemmas. Urban Studies,49(8) 1679–1694.

Neary, J., Egan, M., Keenan, P., Lawson, L. and Bond, L. (2013). Damned if they do,

damned if they don't: Negotiating the tricky context of anti-social behaviour and

keeping safe in disadvantaged urban neighbourhoods. Journal of Youth Studies, 16, pp.

118-134. 10.1080/13676261.2012.710745.

Newton, A. and Hirschfield, A. (2009). Measuring violence in and around licensed premises:

The need for a better evidence base. Crime Prevention and Community Safety, 11, pp.

171-188. DOI: 10.1057/cpcs.2009.12.

Newton, A., Hirschfield, A., Sharratt, K. and Rogerson, M. (2010). Building an evidence base

on alcohol supply points: A pilot project to generate intelligence for managing areas

with licensed premises. Huddersfield: The Applied Criminology Centre, The University

of Huddersfield.

Office for National Statistics. (2004a). Census 2001 key statistics and univariate tables for

Super Output Areas. Titchfield: ONS. Crown Copyright 2004 - Crown Copyright

material is reproduced with the permission of the Controller of HMSO.

Office for National Statistics. (2004b). Super Output Area boundaries. Titchfield: ONS.

Crown copyright 2004 - Crown copyright material is reproduced with the permission of

the Controller of HMSO.

Office for National Statistics. (2010a). Population turnover rates, mid-2008 to mid-2009.

Downloaded from http://www.neighbourhood.statistics.gov.uk/dissemination/Download1.d

Office for National Statistics. (2010b). Mid-2009 population estimates for Middle Layer

Super Output Areas in England and Wales by quinary age and sex. Downloaded from

http://www.ons.gov.uk/ons/publications/re-reference-tables.html?edition=tcm%3A77-

213438.

Office for National Statistics. (2012). Super Output Areas. Downloaded from http://www.ons.gov.uk/ons/guide-method/geography/beginner-s-guide/census/super-

output-areas--soas-/index.html.

Openshaw, S. (1984). The modifiable areal unit problem (concepts and techniques in modern

geography 38) Norwich: GeoBooks.

Ordnance Survey. (2011a). OSMastermap Address Layer 2 data for England. Ordnance

Survey © Crown Copyright. All rights reserved.

Ordnance Survey. (2011b). OS MasterMap address layer and address layer 2 user guide.

Southampton: Ordnance Survey.

Pearce, J., Barnett, R. and Moon, G. (2012). Sociospatial inequalities in health-related

behaviours: Pathways linking place and smoking. Progress in Human Geography, 36,

pp. 3-24. DOI: 10.1177/0309132511402710.

police.uk website. (2014). How are crime locations anonymised? Downloaded from

http://data.police.uk/about/#location-anonymisation.

Rasbash, J., Charlton, C., Browne, W., Healy, M. and Cameron, B. (2009). MLwiN version

2.1. Bristol: Centre for Multilevel Modelling, University of Bristol.

Sampson, R. (2009). Disparity and diversity in the contemporary city: Social (dis)order

revisited. British Journal of Sociology, 60, pp. 1-31. DOI: 10.1111/j.1468-

4446.2009.01211.x.

Sampson, R. and Raudenbush, S. (2004). Seeing disorder: Neighborhood stigma and the

social construction of broken windows. Social Psychology Quarterly, 67, pp. 319-342.

Shaw, C. and McKay, H. (1942). Juvenile delinquency and urban areas. Chicago: University

of Chicago Press.

Simpson, E. (1949). Measurement of diversity. Nature, 163, pp. 688.

Skogan, W. (1990). Disorder and decline: Crime and the spiral of decay in American

neighbourhoods. New York: Free Press.

Smith, D. and Crooks, A. (2010). From buildings to cities: Techniques for the multi-scale

analysis of urban form and function (working paper series 155). London: Centre for

Advanced Spatial Analysis, UCL.

Steptoe, A. and Feldman, P. (2001). Neighborhood problems as sources of chronic stress:

Development of a measure of neighborhood problems, and associations with

socioeconomic status and health. Annals of Behavioral Medicine, 23, pp. 177-185. DOI:

10.1207/S15324796ABM2303_5.

Taylor, J., Twigg, L. and Mohan, J. (2010). Investigating perceptions of antisocial behaviour

and neighbourhood ethnic heterogeneity in the British Crime Survey. Transactions of

the Institute of British Geographers, 35, pp. 59-75. DOI: 10.1111/j.1475-

5661.2009.00365.x.

Taylor, R. (2001). Breaking away from broken windows: Baltimore neighborhoods and the

nationwide fight against crime, grime, fear, and decline. Oxford: Westview Press.

The Countryside Agency, Department for Environment Food and Rural Affairs, Office of the

Deputy Prime Minister, Office for National Statistics and Welsh Assembly

Government. (2004). Rural and urban area classification 2004 an introductory guide.

London: Office for National Statistics.

Tseloni, A. (2007). Fear of crime, perceived disorders and property crime: a multivariate

analysis at the area level. Imagination for Crime Prevention: Essays in Honour of Ken

Pease, Crime Prevention Studies, 21, pp. 163-185.

United Nations Committee on the Rights of the Child. (2008). Forty-ninth session:

Consideration of reports submitted by states parties under article 44 of the convention -

concluding observations: United Kingdom of Great Britain and Northern Ireland.

Geneva: United Nations.

Walker, A., Flatley, J., Kershaw, C. and Moon, D. (2009). Crime in England and Wales

2008/09 volume 1 findings from the British Crime Survey and police recorded crime.

London: Home Office.

Wikström, P. (2009). Questions of perception and reality. The British Journal of Sociology,

60, pp. 59-63. DOI: 10.1111/j.1468-4446.2008.01216.x.

i The seven types of anti-social behaviour have been “teenagers hanging around on the streets”, ”vandalism,

graffiti and other deliberate damage to property or vehicles”, “people using or dealing drugs”, “people being

drunk or rowdy in public places”, “rubbish or litter lying around”, “noisy neighbours or loud parties”, and

“abandoned or burnt-out cars”.

ii Index of Multiple Deprivation (IMD) data are only available at the LSOA level, therefore weighted population

averages (based on 2009 mid-year population estimates from the Office for National Statistics (2010a)) were

calculated to aggregate the data up to Middle Layer Super Output Areas.

iii To protect the identity and privacy of individual victims the point level dataset refers to one of 750,000

'anonymous' map points with the co-ordinates of the actual crime being replaced with the co-ordinates of the

nearest map point (more information on this process can be found at the police.uk website, 2014).

iv The density of pubs, bars and nightclubs per km2 ranged from zero to 74, with the mean number of

establishments per km2 being 1.15 (with a standard deviation of 2.54).

v Whilst we do not include any covariates in our models that relate to Police Force Areas (PFAs), we retain them

in the hierarchy to reflect the design of the survey which is stratified by PFA with unequal probability of

selection between PFAs.

vi Due to space limitations individual-level independent variables (which with the exception of educational

attainment are the same as those detailed in Table 1) are not included in Table 2 but are available on request.

vii The horizontal axis neighbourhood characteristics are presented over the full extent of their observed range

with the exception of the proportion of the population aged between 15 to 24 variable where the x-axis covers

99 per cent of the observed range. Low density of pubs was characterised as -0.45 standard deviations below the

mean (this translated as no pubs in the MSOA) and high density of pubs was defined as one standard deviation

above the mean.

viii Cross-level interactions were also tested to investigate whether the contextual influence of pubs and clubs

diverged based on individual characteristics such as gender and age – no evidence to support this was found.

ix

Higher level variation fell from 0.71 (0.67 at level two (MSOA) plus 0.05 at level three (PFA)) to 0.33 (0.31 at

level two plus 0.02 at level three).

x An example of this methodology is Johnston et al.,’s (2004) analysis of political party support based on the

British Household Panel survey.